-
-
Notifications
You must be signed in to change notification settings - Fork 16
code
PiRadar code This code generates and decodes PiRadar waveforms.
Red Pitaya GNURadio setup procedure and example code
Example of 10 kHz bandwidth phase modulation spectrum for PiRadar, sample rate 100 kHzWe want data to be relayed to the Internet and a cloud server for archival purposes and friendly presenation of data. For our project, I think just getting to the point where the internet-connected nodes shove data into a cloud server, even via SFTP if time runs short is fine.
Given the limited data bandwidth of internode communications ~ 10kbit/s, we will need to do fog computing--preprocess data at the nodes themselves. If we suppose the receiver operates at 100 kS/s and 16 bit/S, we have a 1.6 Mbit/s stream -- far too much to relay across an intermittant HF over the horizon data link.
Suppose we process that down to (quantity / range cell), say there are 50 range cells to each of 8 neighbor nodes, and each measurement quantity is a 32-bit float. Assuming no time averaging, that would yield 8 * 50 * 32 = 12.8 kbit/s. OK, so that's getting close to feasible. We do have to account for RF channel overhead like FEC, but suppose the data is compressible by a factor of 2:1 (or that we time average by a factor of N=2). Voilà, now we have a manageable dataflow.
Oops, we need to consider there may be extended duration comms outages during major solar storms (which is exactly the time that measurements are most precious). We will have to cache processed results for as much as 3 days due to diurnal variations and solar storms.
On-site archival storage is beneficial, since we may later realize there is better processing that can be done, or for some fantastic event we just want all the data we can get.
assuming 200 kB/sec in table below
medium | storage [GB] | storage [days] | cost [USD] | cost/day [USD] |
---|---|---|---|---|
micro SD | 200 | 11 | 70 | 6.36 |
USB HDD | 8000 | 462 | 250 | 0.54 |
Thus with trivial data reduction, each site could go years between human visits with an HDD.
Let's assume 1000 complete writes for simplicity. That's an 11*1000/365 ~ 30 year SD card lifetime. So a potential CONOPS is to write all data to HDD and delete uninteresting times automatically.